CN103492962B

CN103492962B - Rejection of Cyclic Disturbances equipment and Rejection of Cyclic Disturbances method

Info

Publication number: CN103492962B
Application number: CN201280018532.7A
Authority: CN
Inventors: 山口崇; 只野裕吾; 大井伸; 大井一伸
Original assignee: Meidensha Electric Manufacturing Co Ltd
Current assignee: Meidensha Electric Manufacturing Co Ltd
Priority date: 2011-04-15
Filing date: 2012-04-11
Publication date: 2016-11-02
Anticipated expiration: 2032-04-11
Also published as: WO2012141190A1; KR20140002762A; US20140039694A1; US9400508B2; CN103492962A; KR101502020B1

Abstract

In the torque ripple minimization affected by periodic disturbance observer controls, it is necessary to consider that the equipment variation caused by using for a long time and device characteristics change.Therefore, require to improve the robustness to described identification model error always.In order to meet this demand, it is provided that following system, comprising: phasing amount calculating section (11), derive phasing amount by the phase place of the vector locus of the frequency component drafting of basis periodic disturbance；Gain calibration amount calculating section (12), derives gain calibration amount by the gait of march of the vector locus of the frequency component drafting of basis periodic disturbance, and the gait of march calculated is compared with threshold value；And, rotating vector calculating section (13), by the guiding system identification model corrected value that phasing amount is multiplied with gain calibration amount, and carry out the system identification model of calibration cycle disturbance observer part (14) based on described system identification model corrected value.

Description

Rejection of Cyclic Disturbances equipment and Rejection of Cyclic Disturbances method

Technical field

The present invention relates to for suppressing the periodicity of the periodic disturbance produced by control object to disturb Dynamic suppression equipment and suppressing method.

Background technology

When the control equipment and electric machinery control device with learning functionality runs, control object has Time produce periodic disturbance.

In the case of such as motor is as the control object of generation periodic disturbance, produces and claimed For the pulsation (pulsation) of torque ripple (torque ripple), it causes various problems, Such as vibration, noise, take uncomfortable, machine resonance etc..Exactly, it is being widely used Magnet embedded type synchronous motor (being hereinafter referred to as PM motor) in, intricately produce teeth groove Torque ripple and reluctance torque fluctuation.In order to suppress this generation, by electronically to control Object processed applies to compensate signal and suppresses the multiple method of torque ripple to be known to offset pulsation 's.

These methods can be roughly classified into two kinds, and one is feedforward compensation method, and it is based on right The analysis result of electromagnetic field and approximate expression compensate torque ripple, and another kind of for feedback side Method.

Feedback method be grouped into carry out learning by torque meter method, by according to current of electric Fluctuation Assumption torque fluctuation carries out the method for suppression control and based on the current value detected The torque ripple disturbance observer method that suppression controls is carried out with tachometer value.Although these feedbacks Method can tackle wave characteristic variation onlinely, but these methods have according to current fluctuation During Assumption torque fluctuation the error of approximation that is inevitably generated and when high frequency band works by The frequency band that disturbance observer wave filter presents limits.

Torque ripple is the periodic disturbance produced by the rotation change degree of motor, even and if When motor operates with lower speed, the higher order component of periodic disturbance also tends to humorous with machine Vibration frequency mates.So, in the feelings of the variable speed drive system Applied Learning control to oscillatory type Under condition, it is necessary to be considered for the reply of the wild effect caused by amplitude jump and phasing back Measure.To achieve it, typically require by means of complicated high-order model and to this mould Suitable control/the adjustment of type, therefore effective to the periodic disturbance of such as torque ripple etc Control is difficult.For having suppressed the measure of such periodic disturbance according to patent documentation-1 Know.

In the measure of patent documentation 1, as seen from Figure 18, torque command Tref is transfused to To bid value conversion portion 1 with produce the Tong Bu rotating coordinate system of rotation with motor d axle and Current order id* and iq* of q axle (orthogonal dq axle), and these current orders It is imported into phase inverter 2(vector majorization equipment) to stand vector majorization.Based on current order Id* and iq*, phase inverter 2 produces and exports to control PM motor 3, and this PM motor 3 passes through Axle is connected to load.The shaft torque detected value detected by the torque detection installed on axle Tdet and the rotor phase angle θ detected by the location detecting apparatus installed on axle is imported into sight Survey in device part 4.

In observer part 4, the cycle pulsation of PM motor 3 is by the frequency of Fourier transformation Component extraction device (Fourier transformation part FT) is detected as DC current, and in frequency component Periodic disturbance compensated part 4a by periodic disturbance observer and estimate, and estimated value is to press down The mode of periodic disturbance processed is added to current order iq*.

By being n rank (or dimension) sin/cos ripple generating portion represented by SC in Figure 18, And Tr represents that change-over period and n represent compensation order (or degree).

Periodic disturbance observer compensates part 4a and is used as the control device of suppression periodic disturbance One of, and the essential structure compensating part 4a is identical with conventional disturbance observer, and portion Divide 4a control disturbance component dividually.By using, each frequency component is presented complex vector System identification model also compensates as the inverse system model of disturbance observer, direct estimation and to control The disturbance of frequency.By means of which, although it is the most relatively easy to control structure, but No matter being how order can obtain for expected frequence and significantly control effect.

Prior art literature

Patent documentation

Patent documentation 1: international publication WO2010/24195A1

Non-patent literature

Non-patent literature 1:Y.Tadano, et al. " Periodic Learning Suppression Control of Torque Ripple Utilizing System Identification for Permanent Magnet Synchronous Motors”,IEEE IPEC-Sapporo,pp. 1363-1370(2010).

Summary of the invention

The problem to be solved in the present invention

Figure 19 is the control block diagram of the periodic disturbance observer shown in patent documentation 1, its In n rank (dimension) component is shown simply.

P_n: equipment, P^_n: system identification model

r_n: control command

d_n: periodic disturbance, d^_n: periodic disturbance estimated value

y_n: the output of control object

Additional n means n order component.

Above variable is all by X_n=X_An+jX_BnThe complex vector represented.

G_F(S): low pass filter

The object of control object=to be controlled

PDO: periodic disturbance observer (observer of periodic disturbance)

First, system identification had previously been applied to the equipment P as control object_n, and tie Fruit is represented as single order complex vector (1).

P^_n=P^_AN+j P^_Bn... ... ... (1)

It should be noted that, P^_AnIt is the real part of the n order component of recognition result, and P^_BnIt is to identify The imaginary part of the n order component of result.

Such as, by using complex vector to represent and be applied to 1 to 1000Hz for each 1Hz System identification result in the case of, it is possible to constitute and comprise the table of 1,000 single order complex vector elements. It also is able to represent recognition result by approximated equation.In two kinds of measures, system model is always Can represent by using simpler single order complex vector.

It should be noted that, by the P^ appeared below_n、r_n、d_n、d^_nAnd y_nIt is by X_n=X_An +jX_BnThe complex vector represented.

In order to be controlled, equipment output (control object output y_n) by simplifying Fourier's change The low pass filter G changed_F(S), in order to extract the frequency of the control object as periodic disturbance Rate component.Extract frequency component by with the P^ reciprocal being expressed as said system device model_n ^-1 Inverse system be multiplied, control command value r produced by multiplied result and located upstream controller_nIt Between difference be drawn towards adder A1, thus cycle estimator disturbance d_n.At adder A2 from Control command value r_nDeduct periodic disturbance value d^ as the estimation compensating bid value_n, thus The periodic disturbance d of adder A3 is added in suppression to_n.Above-mentioned flow process is to be seen by periodic disturbance Survey the control method of device suppression periodic disturbance.

In this control method for control most important basis be about system identification model The precision of actual value.In order to improve the rejection ability to periodic disturbance, system identification requirement The highest precision.

But, it is difficult to obtain High Definition Systems identification.Namely, it is necessary to take into account due to time long Between use cause equipment variation, to device characteristics variation frequent maintenance etc..Relative to truly The error of value causes until suppression controls to be terminated the prolongation of stabilization time, and in worst In the case of, suppression control self constitutes disturbance component because of phase error, and this causes instability The high likelihood controlled.So, require to improve the robustness identifying model error always.

Thering is provided the present invention is to solve the problems referred to above, and an object of the present invention is to provide The Rejection of Cyclic Disturbances equipment of the error of system identification model can be corrected.

The means of solution problem

In order to solve the problems referred to above, a kind of Rejection of Cyclic Disturbances equipment is characterized by:

Rejection of Cyclic Disturbances device, have at its upstream part at, send control command value Controller, and periodic disturbance observer part, by described periodic disturbance observer portion Divide the output as the control object producing periodic disturbance, the cycle of suppression control object Property disturbance frequency component be multiplied with by the reciprocal inverse system represented of system identification model, thus Cycle estimator disturbance, described periodic disturbance observer part the described periodicity estimated Disturbance is deducted from above-mentioned control command value as compensation bid value, thus suppresses the described cycle Property disturbance；

Phasing device for calculating, is carried out at described Rejection of Cyclic Disturbances device by calculating Described Rejection of Cyclic Disturbances control period, according to each frequency component of described periodic disturbance The phase place of the vector locus drawn in complex vector plane, derives phasing amount；

Gain calibration device for calculating, described defeated by be enter to from described control object Go out, correct the gain of the described output from described control object；And

Rotating vector calculates device, by by from described in described phasing device for calculating Phasing amount is taken mutually with the described gain calibration amount from described gain calibration device for calculating Together, carry out computational system identification model corrected value, and, based on described system identification model school On the occasion of, correct the described system identification model of described Rejection of Cyclic Disturbances device.

Described Rejection of Cyclic Disturbances equipment is to use motor setting as described control object Standby, the output from described motor is imported into described periodic disturbance observer part to estimate Periodic disturbance, the estimated periodicity from described periodic disturbance observer part is disturbed Move and deducted from described control command value, with by inverter controlling institute as compensation bid value State motor, wherein:

Described phasing device for calculating, by being transfused to the described output of described motor, is counted Calculate the phase place of the vector locus drawn according to the described frequency component of described periodic disturbance；

Described gain calibration device for calculating is transfused to the described output of described motor, depends on to calculate According to the gait of march of the described vector locus that the described frequency component of described periodic disturbance is drawn, Described gait of march is compared with threshold value, and, repeat described calculating and compare repeatedly with Calculate gain calibration amount；And

Described rotating vector calculates device by by from described phasing device for calculating Described phasing amount and the described gain calibration amount phase from described gain calibration device for calculating Take together, carry out computational system identification model corrected value, and, based on described system identification mould Type corrected value, corrects the institute of the described periodic disturbance of described periodic disturbance observer part State system identification model.

Described Rejection of Cyclic Disturbances equipment is the equipment that feature is following, by being previously calculated Phasing amount θ gone out^ref _n-1It is added to the product of weight a and anglec of rotation θ, derives by institute State described phasing amount θ that phasing device for calculating provides^ref _n。

A kind of Rejection of Cyclic Disturbances method is characterized by:

The Rejection of Cyclic Disturbances step performed by Rejection of Cyclic Disturbances device, described periodicity Disturbance Rejection device has at part at its upstream, the controller that sends control command value, with And periodic disturbance observer part, will be as generation by described periodic disturbance observer part The frequency of the output of the control object of periodic disturbance, suppression control object periodic disturbance Component is multiplied with by the inverse system represented reciprocal of system identification model, thus cycle estimator is disturbed Dynamic, described periodic disturbance observer part the described periodic disturbance estimated is as compensation Bid value deducts from above-mentioned control command value, thus suppresses described periodic disturbance；

The phasing amount calculation procedure performed by phasing device for calculating, by calculating Described Rejection of Cyclic Disturbances device carries out described Rejection of Cyclic Disturbances control period, according to institute State the phase of the vector locus that each frequency component of periodic disturbance is drawn in complex vector plane Position, derives phasing amount；

The gain calibration amount calculation procedure performed by gain calibration device for calculating, by being transfused to From the described output of described control object, correct the increasing of output from described control object Benefit；And

The rotating vector calculation procedure that device performs is calculated, by will be from described by rotating vector The described phasing amount of phasing device for calculating calculates dress with from described gain calibration amount The described gain calibration amount put is taken together mutually, carrys out computational system identification model corrected value, and, Based on described system identification model corrected value, correct the institute of described Rejection of Cyclic Disturbances device State system identification model.

Described Rejection of Cyclic Disturbances method is to use motor as the side of described control object Method, the output from described motor is imported into described periodic disturbance observer part to estimate Periodic disturbance, the estimated periodicity from described periodic disturbance observer part is disturbed Move and deducted from described control command value, with by inverter controlling institute as compensation bid value State motor, wherein:

Described phasing amount calculation procedure, by being transfused to the described output of described motor, is counted Calculate the phase place of the vector locus drawn according to the described frequency component of described periodic disturbance；

Described gain calibration amount calculation procedure is transfused to the described output of described motor, depends on to calculate According to the gait of march of the described vector locus that the described frequency component of described periodic disturbance is drawn, Described gait of march is compared with threshold value, and, repeat described calculating and compare repeatedly with Calculate gain calibration amount；And

Described rotating vector calculation procedure is by by from described phasing device for calculating Described phasing amount and the described gain calibration amount phase from described gain calibration device for calculating Take together, carry out computational system identification model corrected value, and, based on described system identification mould Type corrected value, corrects the institute of the described periodic disturbance of described periodic disturbance observer part State system identification model.

Utilize features described above, as the error energy of the system identification model of periodic disturbance observer Enough suitably correcting, therefore, the robustness for the error identifying model can improve.

Described Rejection of Cyclic Disturbances equipment is characterised by, calculates dress at described rotating vector The outlet side put arranges learning functionality part, described rotating vector calculate the system that device is derived Identify that model corrected value is stored in described learning functionality part, to described system identification model The corrected value of error learnt, learn to value be imported into described periodic disturbance observer Part is also multiplied to produce the identification model corrected value after renewal with described identification model corrected value.

Utilize features described above, it is possible to study (storage) is for the mistake identifying model of given frequency The corrected value of difference, thus until having corrected during for the repetitive operation of same frequency Learning time can shorten, or perhaps without learning.

Described Rejection of Cyclic Disturbances equipment is characterised by, identifies model calibration function part It is set to described learning functionality part, described identification model calibration function part repeats in institute State the movement/change of operating point when system identification model corrected value is stored, when described operating point Described mobile end time, control to be unlocked to the suppression of described periodic disturbance, and when described When suppression control completes, while storing described system identification model corrected value, to described week The described suppression control of phase property disturbance is closed.

Utilize features described above, even if system identification has error, it is also possible to by model calibration function Complete to suppress control, and be obtained in that new identification model simultaneously.

Described Rejection of Cyclic Disturbances equipment is characterised by, described learning functionality part is provided with Interpolation correction device, thus described system identification model corrected value can have after learning Optional frequency scope.

Utilize features described above, by making the learning outcome in characteristic frequency have influence on surrounding frequencies model Enclose, stablize when frequency variation and the time needed for suppressing control that learns can shorten.

Described Rejection of Cyclic Disturbances equipment is characterised by, calculates to described phasing amount Device, described gain calibration device for calculating and described rotating vector calculate device and provide institute respectively State the n order frequency component of periodic disturbance.

Utilize features described above, even if experiencing the error estimation of system identification model and the multiple of suppression In the presence of the while of periodic disturbance frequency component, it is also possible to carry out suitable correspondence about them.

Another kind of Rejection of Cyclic Disturbances equipment is characterized by:

Phase calculation device, calculates what the position that detected before a cycle extended to detect The Vector rotation angle of current location is as the phase place caused by the error of described system identification model Error, described phase rotation angle is defined relative to following vector, in described vector, Described Rejection of Cyclic Disturbances device carries out described Rejection of Cyclic Disturbances control period according to described The vector locus that each frequency component of periodic disturbance is drawn in complex vector plane is from described inspection The current location measured moves to origin position；

Gain calculating machinery, according to currency and the described vector rail of the described gain of vector locus The expected value of the described gain of mark, calculates and is caused by the described error of described system identification model Gain error, described vector locus is to carry out described week at described Rejection of Cyclic Disturbances device Phase property Disturbance Rejection control period is plotted according to each frequency component of described periodic disturbance In complex vector plane；And

Corrective command value calculation apparatus, calculates rotating vector as described system identification model Corrective command value, described rotating vector is by being calculated by described phase calculation device Described phase error takes one mutually with the described gain error calculated by described gain calculating machinery Rise and derive,

Wherein, described system identification model is by by being calculated by described corrective command value calculation apparatus The described corrective command value that goes out and correct.

Another kind of Rejection of Cyclic Disturbances method is characterized by:

The Rejection of Cyclic Disturbances step performed by Rejection of Cyclic Disturbances device, described periodicity Disturbance Rejection device has at part at its upstream, the controller that sends control command value, with And periodic disturbance observer part, will be as generation by described periodic disturbance observer part The frequency of the output of the control object of periodic disturbance, suppression control object periodic disturbance Component is multiplied with by the inverse system represented reciprocal of system identification model, thus cycle estimator is disturbed Dynamic, described periodic disturbance observer part the described periodic disturbance estimated is by as mending Repay bid value to deduct from above-mentioned control command value, thus suppress described periodic disturbance；

The phase calculation step performed by phase calculation device, calculates and detected before a cycle The position Vector rotation angle of current location that extends to detect as by described system identification The phase error that the error of model causes, described phase rotation angle is by fixed relative to following vector Justice, in described vector, carries out described periodic disturbance at described Rejection of Cyclic Disturbances device Suppression control period is painted in complex vector plane according to each frequency component of described periodic disturbance The vector locus of system moves to origin position from the described current location detected；

The gain calculation procedure performed by gain calculating machinery, according to the described gain of vector locus Currency and the expected value of described gain of described vector locus, calculate described system identification The gain error of the described error of model, described vector locus is in described Rejection of Cyclic Disturbances Device carries out the described Rejection of Cyclic Disturbances control period each frequency according to described periodic disturbance Rate component is plotted in complex vector plane；

The corrective command value calculation procedure performed by corrective command value calculation apparatus, calculate rotate to Amount is as the corrective command value to described system identification model, and described rotating vector is by will be by The described phase error that described phase calculation device calculates calculates with by described gain calculating machinery The described gain error gone out is taken mutually and is derived together；And

The step of described system identification model is corrected by the described corrective command value calculated.

Utilize features described above, it is possible to the error of correction system identification model, thus owing to accurately setting The effect identifying model put, it is possible to perform preferably to suppress to control.

Described another kind of Rejection of Cyclic Disturbances equipment is characterised by, described phase calculation fills Putting and derive the described anglec of rotation in the following way, described mode is by such as next anglec of rotation Being added to another anglec of rotation following, a described anglec of rotation is to examine relative to before a cycle The vector that the position measured is seen to the current location detected, from a week of described vector locus The anglec of rotation that the position detected before phase extends to origin position, another anglec of rotation described is Relative to the position detected before a cycle to the vector in terms of the current location detected, from The anglec of rotation that the current location detected extends to origin position.

Utilize features described above, calculate by system owing to considering past state and current state Identify the phase error caused by error of model, it is possible to improve correction accuracy.

Described another kind of Rejection of Cyclic Disturbances equipment is characterised by, described phase calculation fills Put by the difference between the described anglec of rotation and the expectation anglec of rotation is passed through PI controller Calculate described phase error, and described gain calculating machinery is by by described in described gain Difference between currency and expected value passes through described PI controller to calculate described gain by mistake Difference.

Utilize features described above, it is possible to increase vector locus is followed ability.

Described another kind of Rejection of Cyclic Disturbances equipment is characterised by, to described phase calculation Device, described gain calculating machinery and described corrective command value calculation apparatus provide described periodicity Frequency component described in several rank of disturbance, and, the corrective command value quilt of frequency component described in every rank Derivation is used for correcting described system identification model.

Utilize features described above, even if existing concurrently with for suppressing or correcting system identification model , multiple frequency component of periodic disturbance time, it is also possible to suppress multistage periodic disturbance.

Advantages of the present invention

(1) according to the present invention, as the mistake of the system identification model of periodic disturbance observer Difference can suitably correct, and therefore, the robustness for the error identifying model can improve.

(2) in accordance with the invention it is possible to learn (storage) identification model for given frequency The corrected value of error, thus until having corrected during for the repetitive operation of same frequency Till learning time can shorten, or perhaps without learning.

(3) according to the present invention, even if system identification comprises error, it is also possible to by model school Orthofunction is set up suppression and is controlled, and is obtained in that the identification model after renewal simultaneously.

(4) according to the present invention, by making the learning outcome in characteristic frequency have influence on frequency around Rate, though the time needed for also being able to stabilization checking control when frequency variation and shortening study.

(5) according to the present invention, though system identification model error estimation and suppression with week Multiple frequency components of phase property disturbance simultaneously in the presence of, it is also possible to easily take appropriate measures.

(6) in accordance with the invention it is possible to correct the error of system identification model, thus essence is provided The identification model really arranged, it is possible to perform suitably to suppress to control.

(7) according to the present invention, due to by means of past state and current state calculate by Phase error caused by system identification model error, it is possible to improve correction accuracy.

(8) in accordance with the invention it is possible to improve and vector locus is followed ability.

Accompanying drawing explanation

Fig. 1 is the block diagram identifying model correction portion of first embodiment.

Fig. 2 is for determining the flow chart of gain calibration time in first embodiment.

Fig. 3 is the block diagram identifying model correction portion of the second embodiment.

Fig. 4 is to illustrate the diagram of complex vector planar obit simulation used in the present invention, it explained by The phase error that system identification model error causes.

Fig. 5 (a) and (b) illustrate the example of vector planar obit simulation, and wherein (a) is system The vector locus chart provided when model is error free is provided, and (b) to be system identification model have The vector locus chart provided during error.

Fig. 6 (a) and (b) illustrate the system identification figure provided in a second embodiment under emulation Table.

Fig. 7 (a) and (b) and (c) illustrate and depict in the second embodiment without learning functionality The chart of the result of emulation.

Fig. 8 (a) and (b) and (c) illustrate to depict in the second embodiment have learning functionality The chart of the result of emulation.

Fig. 9 (a) and (b) and (c) illustrate after the amendment depicting the second embodiment in model The chart of simulation result.

Figure 10 is the stream used in the process suppressing incorrect study in the 3rd embodiment Cheng Tu.

Figure 11 illustrates the block diagram used when detecting many orders at the same time in the fourth embodiment.

Figure 12 (a) and (b) and (c) illustrate for explanation use in the 5th embodiment The chart of interpolation processing.

Figure 13 is the control block diagram of the system identification model corrector used in the sixth embodiment.

Figure 14 is to manifest during the system identification model being shown in and applying the invention to error The chart of vector locus.

Figure 15 is to illustrate the complex vector planar obit simulation for explaining seventh embodiment of the invention principle View.

Figure 16 is the control block diagram of the system identification model corrector used in the 8th embodiment.

Figure 17 illustrates the block diagram of diagram the 9th embodiment.

Figure 18 is the block diagram of conventional Disturbance Rejection equipment.

Figure 19 is the block diagram of the controller of periodic disturbance observer.

Detailed description of the invention

Introduce embodiments of the invention below with reference to the accompanying drawings.It should be noted that the present invention is not It is limited to the embodiment hereinafter introduced.In these embodiments, to system identification model error (i.e. The error of system identification model) carry out the error that learns and correction learning arrives.In following example In, there is situation about controlling as the example explanation torque ripple controlling motor.But, this Bright it is not limited to such control.

First embodiment:

First, as wanting controlled output (or controlled output), will pay close attention to such as at real axis On illustrate n rank pulsation extract component (cosine coefficient) T_AnWith in the imaginary axis, illustrate to pulse in n rank Extract component (sinusoidal coefficients) T_BnComplex vector plane on each frequency of torque ripple of drawing The track of rate component.

Fig. 4 is the diagram illustrating complex vector planar obit simulation.By 100_nInstruction is to illustrate from opening Begin to have suppressed the elapsed time [t=t_n] time the position (current location detected) that manifests, and by 100_n-1Instruction is to illustrate elapsed time [t=t from starting suppression_n-1] time the position (that manifests The position detected before the individual cycle).

If system identification model does not has error relative to actual value, then vector locus is from the beginning of controlling Point extend to initial point, from such as illustrate vector planar obit simulation Fig. 5 (a) finding, do not exist turn Square fluctuation (periodic disturbance).It is, vector locus referred to point-blank with the optimal response time To initial point.As from Fig. 5 (b) finding, if system identification model has error relative to actual value, Vector locus is then forced to draw curve or circular trace, and in a worst case scenario, vector Track points to the direction of infinity.

This embodiment is assumed by n rank compensation current command value is set to 0(zero) support Disappear torque ripple (the periodic disturbance d shown in Figure 13_n).But, bid value is being arranged For being not equal to 0(zero) value in the case of, the position of compensation current shown in vector plane Put corresponding to initial point.

In the present invention, at suppression control period, including gain G in below equation (2)^ref And phase theta^refRotating vector P_n ^refBy according to the above-mentioned information about track continuously Determine, and from equation (3) finding, by by rotating vector and identification model P^_nIt is multiplied, Carry out correctly identified model with the identification model P ' after being updated_n.Model P ' is identified after renewal_nQuilt It is applied in periodic disturbance observer PDO the inverse system identification model used.

P_n ^ref=G^ref·(cosθ^ref+jsinθ^ref)..........(2)

P’_n=P^_n·P_n ^ref......................(3)

Fig. 1 illustrates the block diagram of the correction portion identifying model depicting first embodiment, this school Positive part realizes equation (2) and (3).Fig. 1 illustrates in Figure 19 and frequency in a simple manner decoupled The part that component is relevant, and part identical with the part of Figure 19 in Fig. 1 is by identical accompanying drawing Labelling indicates.

In the block diagram of the correction portion of system identification model, by 10 instructions is real equipment, 11 is the phasing amount calculating section of the phasing device for calculating as the present invention, this phase Bit correction amount calculating section is according to the output y of the control object of equipment 10_n, basis is periodically Phase place (anglec of rotation) θ of the vector locus that each frequency component of disturbance is drawn, and just As hereinafter by introduction, calculated correcting value θ^ref。

By 12 instructions is the gain calibration amount of the gain calibration device for calculating as the present invention Calculating section, this gain calibration amount calculating section is according to the output y of control object_n, basis The gait of march of the vector locus that each frequency component of periodic disturbance is drawn (the most currently increases Benefit value) | v |, and flow chart based on Fig. 2 corrects this gait of march.

By 13 instruction be as the present invention rotating vector calculate device rotating vector calculate Part, this rotating vector calculating section is based on equation (2), by by θ^refWith gain G^refPhase Take advantage of, calculate rotating vector P_n ^ref(the corrective command value to system identification model), and pass through Rotating vector P_n ^refThe system identification model of calibration cycle disturbance observer part 14.Control The output y of object_nAll it is input in periodic disturbance observer part 14 with control command, and And the calculating carrying out being described by the block diagram of Figure 13 is to derive periodic disturbance estimated value D^_n, and And in adder A2 from control command r_nIn deduct periodic disturbance estimated value D^_nExist with suppression The periodic disturbance d added at adder A3_n。

Phasing amount θ based on equation (2)^refDetermine in the following manner.

As shown in Figure 4, point to initial point and be placed on the elapsed time from starting to control the time [t=t_n] defined location (T afterwards_An, T_Bn) vector will be designated as P_t, and from time [t=t_n-1] To time [t=t_n] vector will be designated as v.From P_tThe anglec of rotation of the v seen will be designated as θ. If there is no model error, then θ always 0(zero).When phasing amount θ^refIt is reversed to During the phase theta detected, it becomes possible to the error of correction system identification model.

θ=tan^-1(P_t×v/P_t·v)................(4)

In equation (4), [×] represent that apposition and [] represent inner product.

If model error is little to the degree that just can carry out suppressing without the help of correction, then cycle Property disturbance observer part produce compensation order on the direction of θ=0.If there is any mould Type error, then determine correcting value be similar to the phase error of identification model in consideration value θ in the case of. In order to alleviate by with the impact caused by the similarity of phase error, the phasing amount of the present invention Calculating section 11 is by by previous phasing amount θ^ref _n-1With weight a and anglec of rotation θ Product addition derives θ^ref, as shown in equation (5).

It should be noted that, phasing amount θ^ref _nInitial value be 0(zero).This is utilized to process, Even if also be able to suitably adjusted value θ when changing suddenly^ref, thus mode is by advance flexibly Phase, and it is thus possible on the direction making θ be zero calibration model.

θ^ref _n=θ^ref _n-1–a·θ…………………(5)

Gain calibration amount calculating section 12 flow chart based on Fig. 2 determines gain after correction G^ref.Even if in phase error hour, if gain error is big, then also tend to appearance impact point Vibration action and very long convergence time as center.By carrying out gain calibration and phase Bit correction, it is possible to set up the Corrective control sane to model error height.

By means of determining the flow chart of Fig. 2 of gain after correction, according to the position vector of Fig. 4 P_tGain G after correction is determined with velocity vector v^ref.Fig. 2 step from start to finish and suppression control System is repeated cyclically, concurrently so gain calibration amount is by succeedingly (one after another) Determine.About the adjustment direction of correcting value, gain G after correction^refIncrease/reduction corresponding to row Enter the reduction/raising of speed.Gain G after correction^refInitial value be arranged to 1(mono-).

(A) processing procedure to (E) of Fig. 2 is explained below.

(A) if absolute value | r | of position vector is equal to or less than threshold value rth, then judged Convergence state occurs so not being adjusted.

(B) in order to prevent fierce behavior, with the relatively short cycle, (such as, every tens [in the least Second]) carry out the control of gait of march.Gait of march | v1 | calculates according to velocity vector, and And because judgment threshold is proportional to | r | so will be it is determined that be k | r |.If gait of march | v1 | Equal to or higher than threshold value, then judge gain shortage so gain G after Jiao Zheng^refIt is increased a1 [%]. It should be noted that, a1 and k is the parameter that can be arranged generally randomly.

(C) cycle period of Corrective control is divided into N section, and with this cycle period (example As, the most several [seconds]) calculate gait of march | v2 |.

(D) if | v2 | is equal to or less than threshold value vth, then judge that gain is too high so after Jiao Zheng Gain G^refIt is reduced a2 [%].By doing that, it is suppressed that the convergence state of delay also improves Gait of march.It should be noted that, vth and a2 is the parameter that can be arranged generally randomly.

(E) gain G after correction^refThe most limited and store end value in memory.

As described above, by using (Y_AN, Y_BN) vector locus and above-mentioned phase place And gain correction approach, identify that model is used equation (3) to correct.

So, in this embodiment, it is possible to suitably correct in periodic disturbance observer The error of the system identification model used such that it is able to improve for system identification model error Robustness.

Second embodiment:

In the above-described first embodiment, it is provided that in given frequency, suitably calibration cycle is disturbed The device of the system identification model of observer part.And in a second embodiment, by arranging Value P after the correction storing final derivation_n ^refMemorizer, study (storage) function is set.

Fig. 3 illustrates the block diagram identifying model correction portion with learning functionality.By 15 instructions Be memorizer, this memorizer by switch 16 input rotating vector calculating sections 13 outputs P_n ^ref, and it is simultaneously entered the angular velocity n ω of PM motor.It should be noted that, memorizer 15 He Switch 16 composition study funtion part.

Rotating vector P after correction of a final proof_n ^refIt is received and is stored in determining in memorizer 15 Shi Kaiguan 16 carries out on/off operation, and the storage timing for rotating vector is due to application Timing when being adequately suppressed periodic disturbance to the correction process identifying model or time Between.By in such timing rotating vector P_n ^refIt is stored in memorizer 15, is applied to The output of periodic disturbance observer part 14 is from P_n ^refChange into P_n ^mem.As may with etc. The below equation (6) that formula (3) is similar is indicated, in periodic disturbance observer 14, for Model P is identified after deriving correction_n', carry out P_n ^memBe multiplied.

P’_n=P^_n·P_n ^ref·P_n ^mem..............(6)

Fig. 6 to Fig. 9 illustrates the result of two kinds of test simulation, a kind of basis being to use learning functionality The test simulation of embodiment and another kind is the test simulation not using learning functionality.Imitative in test In very, rotary speed and torque are maintained under rigid condition (42 [Hz]), (number of poles=4), (30 [Nm]), and to correct the Posterior circle cycle be 20 [ms].Under the conditions of torque ripple, right Electricity frequency once apply a certain amount of fluctuation with secondary frequency component (1f, 2f).Respectively Gain and Fig. 6 (a) and 6(b of phase place (i.e. analogue system identification diagram) are shown) in, by The instruction of a line is nominal model, b line indicate is variation model 1, and is indicated by c line It it is variation model 2.

For the error condition of system identification model, before carrying out test simulation, Fig. 6 (a) And 6(b) shown in model be arranged to inverse nominal identification model, and by real equipment It is designated as changing model 1 and having carried out error setting.Fig. 7 (a), 7(b) and 7(c) illustrate The situation of disabling learning functionality, and Fig. 8 (a), 8(b) and 8(c) illustrate and enable study merit The situation of energy.In the drawings, the every width of accompanying drawing with additional marking (a) shows that disabling Vector locus in the case of learning functionality, the every width of accompanying drawing with additional marking (b) shows that Away from initial point (T_an), (T_bn) distance | r | and relation between the time, and with additional The every width of accompanying drawing of labelling (c) shows that the relation between shaft torque and time.

As from Fig. 8 (a) and 8(b) it will be appreciated that in nominal identification model, to phase place Error and sensing dissipate the control of the speed in direction and just start after suppression control, thus with Sample in the case of 1f and 2f, do not suppress control dissipate in the case of carry out torque ripple Dynamic suppression.

Fig. 9 (a), 9(b) and 9(c) illustrate system sudden change and disturbance variation checking knot Really.In order to verify, at Fig. 6 (a) and 6(b) describe variation model 1 in complete suppression Under the conditions of, variation model 2 experiences system sudden change, and torque ripple doubles simultaneously.As from Fig. 9 (a), 9(b) and chart finding 9(c), identify that model presents and dash forward for system Change and disturbance change suitably correct and inhibit dissipates action, it is achieved that to torque ripple Dynamic suppression.

So, according to this embodiment, in addition to the effect of first embodiment, also achieve To the study (or storage) for the correcting value identifying model error at certain frequency, so Can shorten until the study period corrected under the repetitive operation at same frequency, Or in extreme circumstances, study may become unnecessary.

3rd embodiment:

In order to learn that Fig. 3 identifying, the correcting value of model, this correcting value are stored in memorizer and (learn Practise funtion part) in.Weight in the gamut of equipment work or while changing operating point The given range of multiple learning manipulation carries out learning manipulation.But, in such learning manipulation phase Between, it is susceptible to learn by mistake.

In the third embodiment, open/close suppression to control so that suppression is when operating point moves The mistake study that can occur.

Figure 10 illustrates the identification model calibration function part institute provided by above-mentioned learning functionality part The process sequence performed.In step S1, operating point is moved, and in step S2, to movement Whether it is complete and judges.It is complete if mobile, then starts suppression control in step S3 System.In step S4, control whether to be complete to suppression and judge, if suppression has controlled Through completing, then store in step S5, and in step S6, close suppression and control.? After, in step S7, whether the movement of all operations point is complete and judges.This sentence Disconnected continue until judged till, so completing the knowledge of whole frequency range after correction Other model.In the case of the most given operating point execution processes sequence, by using any means Derive interval point.

Even if by doing so it is possible, system identification comprises error, it is also possible to by model calibration function Complete to suppress control, and be obtained in that new identification model simultaneously.

4th embodiment:

In the above-described embodiments, by estimating that the identification model error of assigned frequency component makes Can be corrected.In this 4th embodiment, the most as seen from Figure 11, it is individual that " n " is set (order n) the observer part 20 of the frequency component of periodic disturbance₁To 20_N, each have There is identification model correcting unit, in order to be identified model error concurrently for each order Estimate and correction, the estimation such as carried out in the control system of Fig. 1 and Fig. 3 and correction.

There is the corresponding observer part 20 identifying model correcting unit₁To 20_NEach All there is multiple funtion part, the periodic disturbance observation being such as illustrated in Fig. 1 and Fig. 3 Device part 14, phasing amount calculating section 11, gain calibration amount calculating section 12 and rotation Vector calculating section 13, and for n rank control command r₁To r_NEstimate to identify model error And export periodic disturbance estimated value d^ of the result of error correction as system identification model₁ To d^_N。

According to this embodiment, except the advantage that obtains in the first and the second embodiments described above it Outward, there is the advantage obtained further, in the advantage obtained further, even if to be suppressed And the periodic disturbance frequency component being used for estimating system identification model error exists incessantly simultaneously When one, it is also possible to carry out suitable reply action.

5th embodiment:

Although above-mentioned second to the 4th embodiment has adds learning functionality therein to, but learn Practise point for only one dot frequency.

It is, in the frequency response chart of Figure 12, every width all depicts P_A(P_B) pass System, wherein Figure 12 (a) is the P depicting system identification model in the initial state_A(P_B) The frequency response chart of relation.In the above-described embodiments, study identifies the result shape of model error Become as Figure 12 (b) describes, the suppression of periodic disturbance control only for by Assigned frequency point " x " of the variation offer of model or point " y " such that it is able to learning model is by mistake The point of difference is only for assigned frequency.So, as fruit dot only moved little from learning frequency Frequency, then be actually unable in the learning outcome that enough uses are obtained by the frequency being close to this frequency. If control object is the equipment of speed variable, such as motor etc., even if then frequency variation is very Trickle being also required to learns again.

So, in the 5th embodiment, above-mentioned learning functionality part is provided with interpolation correction device, As Figure 12 (c) is described, by interpolation correction device to study point " x " and the frequency of " y " Rate district application interpolation processing, so as from interpolation district " z " finding, it is possible to relative to Accurate Points (pin Point) certain frequency range is obtained.

So, in this embodiment, by making the learning outcome obtained at assigned frequency affect Surrounding frequencies, it is possible to needed for realizing the stability of suppression control and carrying out frequency variation learning The shortening of time.

Sixth embodiment:

Indicate by the reference 30 in Figure 13 is anglec of rotation calculating section, and it is based on control The output y of object processed_n, the vectorial rail of each frequency component drafting of basis periodic disturbance Phase place (anglec of rotation) θ of mark, reference 31 indicate is sign-inverted device, should Sign-inverted device, by above-mentioned phase theta being inverted, derives phase theta^ref(by system identification mould The phase error that type error causes).Anglec of rotation calculating section 30 and sign-inverted device 31 Constitute the phase calculation device of the present invention.

Indicate by reference 40 is present speed calculating section, and it is based on control object Output y_n, vector locus that each frequency component of basis periodic disturbance is drawn current Speed (current gain value) | v |, reference 41 indicate is desired speed calculating device, This desired speed calculating device determines desired speed v^ref, by reference 42 indicate be calculate Device, this calculating device is based on above-mentioned | v | and v^refBetween ratio, calculate gain G^ref(=| v |/v^ref: Gain error by caused by system identification model error).Present speed calculating section 40, phase Hope that speed calculation device 41 and calculating device 42 constitute the gain calculating machinery of the present invention.

Indicate by reference 50 is rotating vector calculating section, and it is by above-mentioned phase place θ^refWith gain G^refTake together mutually, calculate rotating vector P^ref, with by using rotating vector P^refCarry out the system identification model of calibration cycle disturbance observer PDO.This anglec of rotation meter Calculate part and constitute the corrective command value calculation apparatus of the present invention.

From Fig. 4 it will be appreciated that phase theta^ref.It is, in the accompanying drawings, start from suppression control Elapsed time [t=t_n] time provide position 100_n(T_An、T_Bn) vector in initial point direction by P_tInstruction, and from time [t=t_n-1] place provide position 100_n-1To at time [t=t_n] place's offer Position 100_nVector indicated by v.From P_tThe anglec of rotation of v acquired when seeing is by θ table Show, and the phase of this anglec of rotation actual value of being considered as relative system identification model approx Position error.Will be at time [t=t in consideration of it, use equation (7) to derive_n] place's correction Angle, θ^ref。

[expression formula 1]

θ^ref=-θ=-tan^-1(Pt×v)/(Pt·v)..............(7)

In equation (7), [×] mean apposition and [] means inner product.

In order to derive G^ref, by desired speed calculating device 21 based on P_tSize (from initial point Distance) determine desired speed v^ref, and by using computer 22 to derive from equation (8) Value G as the speed relative to present speed | v |^ref。

[expression formula 2]

G^ref=|v|/v^ref...........................(8)

This value will be considered by the gain error caused by system identification model error.

In the present invention, once it is determined that desired speed v^ref, be based on determined by desired speed Correction system identification model such that it is able to obtain desired response.About desired speed, the most just It is that the response speed of suppression control or the shape of response ripple, they are the most always set to ideal Value, because it is difficult to carry out application controls with it.But, even if in this case, pass through Determine that the desired speed realizing requiring also is able to obtain satisfied Expected Response.

Figure 14 is shown below simulation result, and wherein, the present invention is applied to Fig. 5 of error The state of (b).Before application of the present invention, track is the curve shown in Fig. 5 (b).No Crossing, after application of the invention, track almost becomes straight line, this demonstrate that the effect of the present invention.

7th embodiment:

By sixth embodiment being provided the improvement in terms of angle correct precision, constitute this 7th Embodiment.Figure 15 is shown through the track in complex vector plane in Fig. 5 to be expanded in the time [t=t_n-2] position 100 acquired by place_n-2(position detected before i.e. two cycle periods) and carry The vector locus of confession.In fig .15, at time [t=t_nPosition 100 acquired by]_nThe angle at place Error is by θ_tnInstruction.In above-mentioned sixth embodiment, θ_tnIt is considered as relative to identification approx The phase error of the actual value of model.More precisely, owing to moving direction is by periodic disturbance Observer determines, so there exist the possibility that at time [t=t_n+1] position acquired by place is (i.e. The position detected after one cycle period), moving direction illustrates relative to having passed through θ^ref The value that the moving direction that is corrected is greater or lesser.Moreover, other are produced due to this correction Error also tends to occur.

In view of case above, in the 7th embodiment, correction accuracy was by actually using Information is gone to be enhanced.

In fig .15, similar with above-mentioned equation (7), calculate from time [t=t_n-1] obtain Position 100 on initial point direction_n-1Vectorial P_t-1The anglec of rotation of the vector v obtained when seeing θt_n-1.The anglec of rotation so calculated can be considered for relative to actual moving direction, To at time [t=t_n-1] position 100 that obtains_n-1The error being corrected.In consideration of it, According to illustrating anglec of rotation θ t_nWith anglec of rotation θ t_n-1The equation (9) of sum, calculating will to Time [t=t_n] position 100 that obtains_nThe angle, θ that place is corrected^ref。

[expression formula 3]

θ^ref=-(θt_n+θt_n-1)

=-{tan^-1(P_t×v)/(P_t·v)+tan^-1(P_t-1×v)/(P_t-1·v)}....(9)

Accordingly, with respect to above-mentioned sixth embodiment, the difference of the 7th embodiment is, replaces Above-mentioned phase place (anglec of rotation) θ, anglec of rotation calculating section 10 calculates θ t_n+θt_n-1, and its He puts consistent with sixth embodiment (Figure 13).

As it has been described above, reflect current correction, energy by the error produced after legacy data is corrected Enough precision improving correction.

8th embodiment:

In above-mentioned 6th and the 7th embodiment, the most true based on current information and past information Determine rotating vector P^ref _nGain G^refAnd phase theta^ref.But, the control device of Figure 16 is used (61,62) also are able to determine gain G^refAnd phase theta^ref。

It is, it is similar with sixth embodiment, for the phase place (anglec of rotation) of vector locus The expected angle of θ and gain v is arranged by expected angle calculating device 63, and desired speed is by the phase Hope that speed calculation device 64 is arranged.Desired speed is expected with mention in sixth embodiment identical Angle is substantially 0(zero).

The expected angle calculated by expected angle calculating device 63 with by anglec of rotation calculating section Difference between the current angular θ that 30 calculate is derived by adder 65, and the difference derived is defeated Enter to PI controller 61, and the desired speed calculated by desired speed calculating device 64 with by work as Difference between present speed | v | that front speed calculating section 40 calculates is derived by adder 66, and And the difference derived is imported into another PI controller 62, to calculate phase theta respectively^refAnd gain G^ref。

If it should be noted that, stem portion identical with Figure 13 in Figure 16 is by identical reference table Show.Above-mentioned PI controller 61 and 62 can be general type, and PID controls to be available 's.

In the 8th embodiment, anglec of rotation calculating section 30 phase place (the anglec of rotation calculated Degree) " θ t can be mentioned in the 7th embodiment_n+θt_n-1" to replace θ.

Utilize above-mentioned layout, compared with above-mentioned 6th and the 7th embodiment, improve vector locus Ability is followed to expected value.

9th embodiment:

In above-mentioned six, the 7th and the 8th embodiment, in assigned frequency component, have estimated knowledge Other model error, and to mention the correction to error be possible.In the 9th embodiment, just As seen from Figure 17, it is provided that " N " individual (order n) PDO of the frequency component of periodic disturbance 70P₁To 70P_N, each there is identification model corrector, in order to parallel for each order Be identified estimation and the correction of model error, such as in the control system of Figure 13 and Figure 16 In the estimation that carries out and correction.

Each there is the PDO70P identifying model corrector₁To 70P_NIn each Have be illustrated in Figure 13 (sixth embodiment) with the function of lower part: periodic disturbance Observer PDO14, anglec of rotation calculating section 30, sign-inverted device 31, present speed Calculating section 40, desired speed calculating device 41, calculating device 42 and rotating vector calculating part Divide 50, or each there is the following functions being illustrated in Figure 16 (the 8th embodiment): Periodic disturbance observer 14, anglec of rotation calculating section 30, present speed calculating section 40, Rotating vector calculating section 50, PI controller 61 and 62, expected angle calculating section 63, phase Hope speed calculating section 64 and adder 65 and 66, in order to be respectively directed to n rank control command r₁To r_NEstimate to identify model error, and export as error based on estimation system identification Periodic disturbance estimated value d^ of the result of model correction₁To d^_N。

Utilize above-mentioned layout, even if to be suppressed and to be used for estimating system identification model error Periodic disturbance frequency component when simultaneously there is more than one, it is also possible to carry out suitable should To action.

Industrial applicibility:

The present invention is applicable to such as to suppress the shaft torque resonance of ergometer system, motor housing Vibration suppression (i.e. for improving the equipment of the ride quality of motor vehicle, elevator etc.), with And other have all devices of the problem about periodic disturbance.

Description of reference numerals

1..... bid value conversion portion

2.... phase inverter

3.....PM motor

4..... observer part

10... real equipment (control object)

11... phasing amount calculating section

12... gain calibration amount calculating section

13,50... rotating vector calculating section

14... periodic disturbance observer part

15... memorizer

16... switch

20₁To 20_n... .. has the PDO of model error correction device

30... anglec of rotation calculating section

31... sign-inverted device

40... present speed calculating section

41... desired speed calculating device

42... calculating device

61,62...PI controller

63... expected angle computer

65,66... adder

70P₁To 70P_N... there is the PDO identifying model corrector

Claims

1. a Rejection of Cyclic Disturbances equipment, has:

Rejection of Cyclic Disturbances device, has in the upstream portion office of control object, input control The controller of bid value processed, and periodic disturbance observer part, described periodic disturbance is seen Survey device part and extract the frequency component corresponding with periodic disturbance from the output of control object, carried The frequency component taken is multiplied with by the inverse system represented reciprocal of system identification model, thus estimates Meter periodic disturbance, estimated periodic disturbance is ordered from above-mentioned control as compensation bid value Make in value and deducting, thus suppress described periodic disturbance；

It is characterized in that also having:

Phasing device for calculating, is worked as by described Rejection of Cyclic Disturbances by input calculating When the described frequency component of the described periodic disturbance that device obtains is drawn in complex vector plane The track provided, derives the phasing amount of described system identification model；

Gain calibration device for calculating, is filled by described Rejection of Cyclic Disturbances by inputting and calculating Put the described frequency component of the described periodic disturbance of acquirement, derive described system identification model Gain calibration amount；And

Rejection of Cyclic Disturbances equipment the most according to claim 1, wherein, it is right to control As if motor, the output from described motor is imported into described periodic disturbance observer part With cycle estimator disturbance, from the estimated week of described periodic disturbance observer part The disturbance of phase property is deducted as compensation bid value, to pass through phase inverter from described control command value Control described motor, wherein:

Described phasing device for calculating, by being transfused to the described output of described motor, is counted Calculate the phase place of the described track drawn according to the described frequency component of described periodic disturbance；

Described gain calibration device for calculating is transfused to the described output of described motor, depends on to calculate According to the gait of march of the described track that the described frequency component of described periodic disturbance is drawn, by institute State gait of march to compare with threshold value, and, repeat described calculating and compare repeatedly to calculate Gain calibration amount；And

Rejection of Cyclic Disturbances equipment the most according to claim 1 and 2, it is further Be characterised by, by by the product addition of weight a and anglec of rotation θ to being previously calculated out Phasing amount θ^ref _n-1, the described phase provided by described phasing device for calculating is provided Bit correction amount θ^ref _n。

Rejection of Cyclic Disturbances equipment the most according to claim 1 and 2, it is further Being characterised by, the outlet side calculating device at described rotating vector arranges learning functionality part, The system identification model corrected value that being calculated device by described rotating vector derives is stored in described Practise in funtion part, the corrected value of the error of described system identification model is learnt, study To value be imported into described periodic disturbance observer part and corrected with described identification model Value is multiplied to produce the identification model corrected value after renewal.

Rejection of Cyclic Disturbances equipment the most according to claim 4, it is the most special Levy and be, identify that model calibration function part is set to described learning functionality part, by described Identify that model calibration function part repeats the behaviour when described system identification model corrected value is stored Make the movement/change of point, when the described mobile end of described operating point, described periodicity is disturbed Dynamic suppression controls to be unlocked, and when described suppression control completes, knows storing described system While other model corrected value, the described suppression control to described periodic disturbance is closed.

Rejection of Cyclic Disturbances equipment the most according to claim 5, it is the most special Levying and be, described learning functionality part is provided with interpolation correction device, thus institute after learning State system identification model corrected value and can have optional frequency scope.

Rejection of Cyclic Disturbances equipment the most according to claim 1 and 2, it is further It is characterised by, to described phasing device for calculating, described gain calibration device for calculating Calculate device with described rotating vector and the n order frequency component of described periodic disturbance is provided respectively.

8. a Rejection of Cyclic Disturbances equipment, has:

Rejection of Cyclic Disturbances device, have the upstream portion office of control object, send control The controller of bid value processed, and periodic disturbance observer part, by described periodic disturbance Observer part using as produce periodic disturbance control object output, suppression control right The frequency component of the periodic disturbance of elephant and the inverse system phase by the expression reciprocal of system identification model Take advantage of, thus cycle estimator disturbance, described periodic disturbance observer the part institute estimated State periodic disturbance to be deducted from above-mentioned control command value as compensation bid value, thus suppress Described periodic disturbance；

It is characterized in that also having:

Phase calculation device, calculates what the position that detected before a cycle extended to detect The Vector rotation angle of current location is as the phase place caused by the error of described system identification model Error, described Vector rotation angle is defined relative to following vector, in described vector, Described Rejection of Cyclic Disturbances device carries out described Rejection of Cyclic Disturbances control period according to described The vector locus that each frequency component of periodic disturbance is drawn in complex vector plane is from described inspection The current location measured moves to origin position；

Gain calculating machinery, according to currency and the described vector rail of the gain of described vector locus The expected value of the described gain of mark, calculates and is caused by the described error of described system identification model Gain error；And

Wherein, described system identification model is by being calculated by described corrective command value calculation apparatus Described corrective command value correct.

Rejection of Cyclic Disturbances equipment the most according to claim 8, it is the most special Levying and be, described phase calculation device derives the described anglec of rotation in the following way, described Mode is that such as next anglec of rotation is added to another anglec of rotation following, a described anglec of rotation It is to the vector in terms of the current location detected relative to the position detected before a cycle, The anglec of rotation that the position detected before a cycle of described vector locus extends to origin position Degree, another anglec of rotation described is to detecting relative to the position detected before a cycle The vector seen of current location, the anglec of rotation extended to origin position from the current location detected Degree.

The most according to claim 8 or claim 9, Rejection of Cyclic Disturbances equipment, it is further It is characterised by, orders to described phase calculation device, described gain calculating machinery and described correction Value calculation apparatus is made to provide frequency component described in several rank of described periodic disturbance, and, every rank The corrective command value of described frequency component is derived for correcting described system identification model.

11. 1 kinds of Rejection of Cyclic Disturbances equipment, have:

It is characterized in that also having:

Phase calculation device, by extending to detect by the position detected before a cycle Current location Vector rotation angle and the Mean Vector anglec of rotation between difference be delivered to PI In controller, calculate the phase error caused by the error of described system identification model, described The position detected before a cycle extends to the Vector rotation angle of the current location detected Defined, in described vector, at described Rejection of Cyclic Disturbances device relative to following vector Carry out described Rejection of Cyclic Disturbances control period to divide according to each frequency of described periodic disturbance The vector locus that amount is drawn in complex vector plane moves to former from the described current location detected Point position；

Gain calculating machinery, by by the currency of the gain of described vector locus and described vector Difference between the expected value of the described gain of track is delivered in described PI controller, calculate by The gain error that the described error of described system identification model causes；And

12. Rejection of Cyclic Disturbances equipment according to claim 11, it is further Being characterised by, described phase calculation device is derived described before a cycle in the following way The position detected extends to the Vector rotation angle of the current location detected, described mode is To be added to another anglec of rotation following such as next anglec of rotation, a described anglec of rotation is relative In the position detected before a cycle to the vector in terms of the current location detected, from described The anglec of rotation that the position detected before one cycle of vector locus extends to origin position, institute It is current to detect relative to the position detected before a cycle for stating another anglec of rotation The vector that position is seen, the anglec of rotation extended to origin position from the current location detected.

13. 1 kinds of Rejection of Cyclic Disturbances methods, have:

The Rejection of Cyclic Disturbances step performed by Rejection of Cyclic Disturbances device, described periodicity Disturbance Rejection device has in the upstream portion office of control object, the control of input control command value Device processed, and periodic disturbance observer part, described periodic disturbance observer part is from control The frequency component corresponding with described periodic disturbance, the frequency extracted are extracted in the output of object processed Component is multiplied with by the inverse system represented reciprocal of system identification model, thus cycle estimator Disturbance, estimated periodic disturbance deducts from above-mentioned control command value as compensating bid value, Thus suppress described periodic disturbance；

It is characterized in that also having:

The phasing amount calculation procedure performed by phasing device for calculating, by input also Described frequency component when the described periodic disturbance obtained by described Rejection of Cyclic Disturbances step The track provided when being drawn in complex vector plane, derives the phase of described system identification model Bit correction amount；

The gain calibration amount calculation procedure performed by gain calibration device for calculating, by input also The described frequency calculating the described periodic disturbance obtained by described Rejection of Cyclic Disturbances step is divided Amount, derives the gain calibration amount of described system identification model；And

The rotating vector calculation procedure that device performs is calculated, by will be from described by rotating vector The described phasing amount of phasing amount calculation procedure calculates step with from described gain calibration amount Rapid described gain calibration amount is taken together mutually, carrys out computational system identification model corrected value, and, Based on described system identification model corrected value, correct the institute of described Rejection of Cyclic Disturbances device State system identification model.

14. Rejection of Cyclic Disturbances methods according to claim 13, wherein, control To liking motor, the output from described motor is imported into described periodic disturbance observer portion Point with cycle estimator disturbance, estimated from described periodic disturbance observer part Periodic disturbance is deducted as compensation bid value, with by anti-phase from described control command value Device controls described motor, wherein:

Described phasing amount calculation procedure makes the described output of described motor to be input to In described phasing device for calculating, with the described frequency of periodic disturbance described in basis The phase place of the described track that component is drawn；

Described gain calibration amount calculation procedure makes the described output of described motor to be input to In described gain calibration device for calculating, with the described frequency of periodic disturbance described in basis The gait of march of the described track that component is drawn, compares described gait of march with threshold value, Further, repeat described calculating and compare repeatedly to calculate gain calibration amount；And

Described rotating vector calculation procedure is by by from described phasing device for calculating Described phasing amount and the described gain calibration amount phase from described gain calibration device for calculating Take together, calculate described system identification model corrected value, and, know based on described system Other model corrected value, corrects the described periodic disturbance of described periodic disturbance observer part Described system identification model.

15. 1 kinds of Rejection of Cyclic Disturbances methods, have:

The Rejection of Cyclic Disturbances step performed by Rejection of Cyclic Disturbances device, described periodicity Disturbance Rejection device has in the upstream portion office of control object, the control that sends control command value Device processed, and periodic disturbance observer part, will by described periodic disturbance observer part As produce periodic disturbance control object output, suppression control object the described cycle Property disturbance frequency component be multiplied with by the reciprocal inverse system represented of system identification model, thus Cycle estimator disturbance, described periodic disturbance observer part the described periodicity estimated Disturbance is deducted from above-mentioned control command value as compensation bid value, thus suppresses the described cycle Property disturbance；

It is characterized in that also having:

The phase calculation step performed by phase calculation device, calculates and detected before a cycle The position Vector rotation angle of current location that extends to detect as by described system identification The phase error that the error of model causes, described Vector rotation angle is by fixed relative to following vector Justice, in described vector, carries out described periodic disturbance at described Rejection of Cyclic Disturbances device Suppression control period is painted in complex vector plane according to each frequency component of described periodic disturbance The vector locus of system moves to origin position from the described current location detected；

The gain calculation procedure performed by gain calculating machinery, according to the gain of described vector locus Currency and the expected value of described gain of described vector locus, calculate and known by described system The gain error that the described error of other model causes；